Three-Dimensional Shape Expression Method and Device Thereof

1. A three-dimensional shape expression method, comprising following steps: extracting a hybrid type framework of a three-dimensional shape; obtaining a segmentation of the three-dimensional shape by segmenting the hybrid type framework; obtaining a sub-structure of the three-dimensional shape according to the segmentation of the three-dimensional shape; and establishing an expression of the three-dimensional shape by using a bag-of-words model according to the sub-structure of the three-dimensional shape.

2. The three-dimensional shape expression method according to claim 1 , wherein the step of extracting the hybrid type framework of the three-dimensional shape comprises: obtaining sampling points by sampling surfaces of the three-dimensional shape; and re-expressing the sampling points to obtain the hybrid type framework comprising a one-dimensional curve and a two-dimensional slice.

3. The three-dimensional shape expression method according to claim 2 , wherein the step of obtaining the segmentation of the three-dimensional shape by segmenting the hybrid type framework comprises: segmenting the hybrid type framework; and obtaining the segmentation of the three-dimensional shape by segmenting the hybrid type framework, according to corresponding relationships between the hybrid type framework and the sampling points.

4. The three-dimensional shape expression method according to claim 1 , wherein the step of obtaining the sub-structure of the three-dimensional shape according to the segmentation of the three-dimensional shape comprises: obtaining a plurality of components of the three-dimensional shape by the segmentation of the three-dimensional shape; establishing a connecting graph connected with the plurality of components of the three-dimensional shape; and extracting a sub-graph of the connecting graph as the sub-structure of the three-dimensional shape.

5. The three-dimensional shape expression method according to claim 1 , wherein the step of establishing the expression of the three-dimensional shape by using the bag-of-words model according to the sub-structure of the three-dimensional shape comprises: matching the sub-structure of the three-dimensional shape with each candidate sub-structure in a candidate sub-structure set to determine a frequency of each candidate sub-structure appearing in the three-dimensional shape; establishing term vectors of the three-dimensional shape according to the frequency of each candidate sub-structure appearing in the three-dimensional shape; and normalizing the term vectors to obtain a bag-of-words expression of the three-dimensional shape.

6. The three-dimensional shape expression method according to claim 5 , wherein before the step of establishing the expression of the three-dimensional shape by using the bag-of-words model, the method further comprises: establishing the candidate sub-structure set; wherein the step of establishing the candidate sub-structure set comprises: obtaining all the sub-structures of the three-dimensional shape in an input data set; determining similarities among the obtained sub-structures of the three-dimensional shape; and selecting the candidate sub-structure from the obtained sub-structures of the three-dimensional shape to form the candidate sub-structure set according to the similarities among the obtained sub-structures of the three-dimensional shape.

7. The three-dimensional shape expression method according to claim 6 , wherein the step of determining the similarities among the obtained sub-structures of the three-dimensional shape comprises: defining a graph kernel among the obtained sub-structures of the three-dimensional shape; and determining the similarities among the obtained sub-structures of the three-dimensional shape according to the graph kernel.

8. The three-dimensional shape expression method according to claim 7 , wherein the step of defining the graph kernel among the obtained sub-structures of the three-dimensional shape comprises: defining a node kernel and an edge kernel; wherein the node kernel is: k node (n i , n j )=1−D(h i , h j )/D h max ; the edge kernel is: k edge (e i , e j )=1−D(u i , u j )/D u max ; k node (n i , n j ) represents the node kernel, k edge (e i , e j ) represents the edge kernel, n i , n j represent the nodes, h i and h j are respectively formed by a connection of a geometric feature histogram of components of the node n i and the node n j , D is a normalized correlation of h i and h j , D h max is the largest value of a distance D(h i , h j ) of any two pairs h i and h j , u i and u j are two-dimensional histograms formed by all points in two connecting components with respect to a vertical upward angle and the distance.

9. A three-dimensional shape expression computer program product, stored in a non-transitory tangible computer-readable medium, comprising: a framework extracting module configured to extract a hybrid type framework of a three-dimensional shape; a segmentation module configured to obtain a segmentation of the three-dimensional shape by segmenting the hybrid type framework; a sub-structure extracting module configured to obtain a sub-structure of the three-dimensional shape according to the segmentation of the three-dimensional shape; and an expression module configured to establish an expression of the three-dimensional shape by using a bag-of-words model according to the sub-structure of the three-dimensional shape.

10. The three-dimensional shape expression computer program product according to claim 9 , wherein the framework extracting module samples surfaces of the three-dimensional shape to obtain sampling points and re-expresses the sampling points to obtain the hybrid type framework comprising a one-dimensional curve and a two-dimensional slice; wherein segmentation module segments the hybrid type framework and obtains the segmentation of the three-dimensional shape by corresponding relationships between the hybrid type framework and the sampling points.